Leveraging Machine Learning for Effective Device Detection and Security in LoRa-Based IoT Systems

LoRa networks, which enable long-range, low-power communication for Internet of Things (IoT) applications, face increasing device identification and security challenges. Detecting and classifying devices in LoRa networks is crucial for mitigating risks such as spoofing and unauthorized transmissions. This study evaluates the performance of three machine learning-based device detection algorithms—Local Outlier Factor (LOF), Autoencoder, and Isolation Forest—using real-world LoRa transmission data. The results indicate that the Autoencoder performs best, achieving an accuracy of 97.50 % with precision, recall, and F1 scores of 0.98, indicating its superior ability to identify devices accurately. The LOF algorithm follows closely with an accuracy of 95.84 %, while the Isolation Forest shows an accuracy of 95.01 %, demonstrating its effectiveness but slightly lower performance in comparison. These findings highlight the potential of machine learning techniques for reliable device detection in LoRa networks, enabling improved security and efficient management of IoT devices.